Wykaz obszarów badawczych:
| # | Obszar badawczy | Dziedzina naukowa |
|---|---|---|
| 1 |
“Machine learning-based diagnosis of spine injuries using computed tomography” Computed tomography (CT) is a crucial imaging technique in medical diagnosis and is the preferred modality for assessing spinal trauma. However, the large volume of image data generated during tomographic examinations presents significant challenges for image analysis and diagnosis. Artificial intelligence (AI) offers the potential to enhance the speed and accuracy of diagnosis in such cases. This research aims to explore the application of machine learning (ML) methods and deep neural networks (DNNs) for the automated detection of traumatic vertebral body injuries. The study will focus on classifying vertebral fractures, distinguishing between traumatic and non-traumatic cases. The learning dataset will be constructed using trauma examination records from a clinical hospital in Warsaw. These records will consist of X-ray spine tomography studies in DICOM format, annotated by experienced radiologists. The AO Spine Classification system for thoracolumbar injuries will serve as the framework for categorizing spinal fractures. To address the computational challenges posed by the high dimensionality of tomographic data (i.e., the large number of voxels per examination), methods for reducing data size will be employed. Various deep neural network architectures will be evaluated to determine their efficacy and performance in fracture classification. Furthermore, the interpretability and explainability of the developed ML-based approach will be analyzed using tools and techniques from Explainable AI (XAI). Efforts will be made to validate the reliability of the recommendations generated by the ML models, ensuring they align with clinical expertise and established diagnostic standards.
|
Inżynieria Biomedyczna |
| 2 |
“Deep neural networks for image reconstruction in electrical capacitance tomography” Electrical capacitance tomography (ECT) enables the visualization of the spatial distribution of an object’s electrical permittivity. Image reconstruction in ECT presents a significant challenge as it is an ill-posed and ill-conditioned inverse problem. Advanced nonlinear algorithms, such as the Levenberg-Marquardt method, are iterative and computationally intensive, primarily due to the repeated calculation of the Jacobian matrix. Deep neural networks (DNNs) have emerged as a promising alternative for image reconstruction. This work will explore deep network architectures that not only match but significantly outperform classical methods in reconstruction quality. Synthetic data will be employed for supervised learning, and the performance of DNNs will be tested against real measurements. Application-specific training datasets will be analyzed, and their size will be expanded using data augmentation techniques. A comparative evaluation of various training datasets will also be conducted. To optimize the network, different loss functions and solvers will be utilized. Reconstructed images will be generated using DNNs and compared against a baseline provided by the Levenberg-Marquardt algorithm. Results will be assessed using selected image quality metrics. It is anticipated that deep networks will enhance the spatial resolution of ECT scanners. The insights gained from selecting deep networks, analyzing their architectures, and refining training strategies can potentially be applied to solve inverse problems in other fields. |
Inżynieria Biomedyczna |
| 3 |
“Capacitively coupled impedance tomography for anatomical and functional imaging” To date, electrical impedance tomography (EIT) using sinusoidal excitation has been regarded as the most promising electrical imaging technique for diagnostic medical applications. However, the high impedance at the electrode-skin contact remains a significant challenge, hindering both the development and practical implementation of this technology. To address these limitations, an alternative approach utilizing non-contact electrodes and pulse excitation will be explored. This method will involve signal shape analysis to determine both components of admittance. Such data will enable the reconstruction of images depicting electrical permittivity and conductivity. Capacitively coupled electrical tomography will be investigated using numerical and physical lung phantoms, with a focus on regional ventilation distribution. Key metrics such as measurement sensitivity, contrast, and spatial-temporal resolution of the resulting images will be evaluated. Non-linear iterative algorithms and deep learning techniques will be employed for image reconstruction. Real measurements will be conducted using a simplified thorax phantom designed to simulate the respiratory cycle. A prototype flexible sensor incorporating surface electrodes will be developed, along with a mechanical-electrical lung phantom. Measurements will be carried out using the 32-channel electrical capacitance tomograph EVT4, which was designed and constructed at ZEJiM.
|
Inżynieria Biomedyczna |
| 4 |
Synthetic Data Generation for Machine Learning-Based Computer-Aided Diagnosis in Spinal Computed Tomography Machine learning-based computer-aided diagnosis in computed tomography (CT) is often limited by the scarcity of large, well-annotated clinical datasets. Acquiring representative CT examinations of vertebral fractures with expert annotations is particularly challenging because many clinically important fracture types occur infrequently, leading to severe class imbalance and reduced model generalization. This doctoral research aims to develop and evaluate methodologies for generating realistic synthetic three-dimensional CT data to support automated vertebral fracture classification. The work will focus on supervised deep learning models implementing the nine-class AO Spine fracture classification system. The research will investigate multiple approaches to synthetic volumetric CT generation, including generative artificial intelligence models and anatomically constrained non-rigid transformations derived from clinical CT examinations. A reference dataset of expert-annotated clinical CT scans will be used for model development, parameter optimization, and independent validation. A central objective is to quantify the effect of different synthetic data generation strategies on the performance, robustness, calibration, and generalization of deep learning models. Classifiers trained on real, synthetic, and hybrid datasets will be systematically compared using independent clinical CT examinations excluded from the training process. Particular attention will be given to improving the recognition of underrepresented fracture classes while maintaining high diagnostic performance across all AO Spine categories. The expected scientific contribution is the development of methodologies for generating anatomically realistic synthetic CT data and a comprehensive evaluation of their effectiveness in mitigating data scarcity in machine learning for computer-aided diagnosis.The research is expected to establish methodological principles for integrating synthetic data into medical imaging workflows and to improve the reliability and generalizability of automated vertebral fracture classification.
|
Inżynieria Biomedyczna |
| 5 |
Development of Explainability Methods for Localizing Vertebral Fractures in Machine Learning-Based Computer-Aided Diagnosis Using Computed Tomography Machine learning-based computer-aided diagnosis in computed tomography (CT) has demonstrated promising performance in automated vertebral fracture classification. However, most supervised learning models are trained using only examination-level or vertebra-level labels and therefore provide little information about the image regions responsible for their predictions. This lack of interpretability limits clinical confidence and hinders the adoption of artificial intelligence in routine radiological practice. Pixel-level annotation of fracture regions could address this issue but is prohibitively time-consuming and expensive because it requires extensive expert involvement. This doctoral research aims to develop explainability methods capable of accurately localizing vertebral fractures using only classification labels during model training. The study will investigate whether explainability maps generated by deep learning classifiers can be transformed into reliable indicators of pathological regions without requiring manually annotated segmentation masks. The research will focus on automated classification of vertebral fractures according to the nine-class AO Spine classification system using three-dimensional CT data. Several deep learning architectures and explainability techniques will be investigated, including gradient-based attribution methods, activation map approaches, and perturbation-based explanations. Novel methodologies will be developed to improve the localization accuracy, anatomical consistency, and clinical interpretability of explainability maps. The proposed methods will be evaluated using expert-annotated clinical CT examinations, with particular emphasis on their ability to identify fracture regions while preserving high classification performance. The expected scientific contribution is the development of explainability methodologies that bridge the gap between image-level classification and lesion localization in medical imaging.
|
Inżynieria Biomedyczna |